{"id":"357a9ad6-c1e1-48c5-8304-38e91eed2930","arxiv_id":"2608.08863","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Homothetic hyperboloidal coordinates give semilinear wave tails the same exponential decay rate at every compactified radius, removing the late-time resolution bottleneck.","lead":"Late-time wave tails decay at different speeds at infinity and at finite radius, and that split makes long numerical simulations expensive. This paper introduces homothetic hyperboloidal coordinates that keep the tail profile smooth everywhere, so simulations reach very late times in far fewer steps.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Uniform-exponential claim rests on unproved self-similar ansatz (63)-(65); numerical confirmation shown only for cubic monopole, leaving higher-multipole rates and interior profiles unverified.","rationale":"The paper is honest about the conditional nature of the analytic step: Sec. V C starts from 'we assume that a leading asymptotic representation exists.' That assumption is exactly what converts power-law tails in u into exponential profiles in tau. The mechanism is plausible, and the spherical mode is supported by prior small-data results; the balance-law tests and spectral convergence provide additional independent evidence. My concern is not that the authors are wrong, but that the central efficiency claim for l>=1 is supported only by I+-extracted rates in Table I and by Fig. 4 for the l=0 cubic case. The p=7 rows being 5-6% above the predicted value suggests the fit interval may be pre-asymptotic, which raises the stakes for a direct multi-radius profile test. The derivative estimate (65) is essential to Eq. (70) and is not an obvious consequence of the pointwise bound (56). A single targeted run checking profile stationarity and rate agreement at several fixed rho for l>=1 would settle whether the concern is real. Until then, conditional acceptance is the right verdict.","tokens_in":23198,"tokens_out":7242,"duration_ms":73997,"concrete_test":"Evolve l=1 and l=2 modes for p=5 (and p=7, if feasible) with N_r=256 to tau=10 in homothetic coordinates. At rho=0.2, 0.5, 0.8, 1.0 compute E(tau,rho)=exp(qtilde_l tau) psi_l(tau,rho) and its tau-derivative. If (a) E(tau,rho) converges to a nonzero constant profile at each radius, (b) d_tau ln E tends to 0 at an exponential rate over the last several units of tau, and (c) the local rates -d_tau ln|psi_l| agree at all four radii to the same tolerance quoted for the monopole (spread <= 1e-5), then the uniform-profile mechanism is confirmed for l>=1. If the profiles keep drifting or interior radii show a different rate, the central claim lacks numerical support beyond l=0.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—same exponential rate at every fixed compactified radius—is derived in Sec. V C from the assumed leading representation (63)-(65): chi_l(u,rho)=u^{-q}F_l(rho)+R_l, with origin-regular F_l and uniform remainder bounds including u dR_l/du = O(u^{-q-delta} rho^{l+1}). The paper explicitly flags this as an assumption, and it is load-bearing: Eq. (70), which converts exponential tau-decay into a uniform local rate, uses the derivative bound to control d_tau epsilon_l. For l=0, the cited small-data estimate (56) gives the leading power, but I do not see that it supplies the uniform derivative remainder estimate (65); for l>=1, no proof of either the leading representation or the remainder is available. The only direct numerical check of the uniform profile is Fig. 4, which is the cubic monopole; Table I reports rates only at rho=1. Thus the efficiency claim for higher multipoles rests on an unverified ansatz, not on measured convergence at multiple radii.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript introduces homothetic hyperboloidal coordinates for the numerical evolution of semilinear wave equations in Minkowski spacetime. It argues that in these coordinates late-time tails take the form chi_l(tau,rho) ~ e^{-q tau} P_l(rho) with the same decay exponent at every fixed compactified radius, so the radial profile remains smooth and late retarded times are reached in a number of time steps that grows logarithmically with u. The authors implement the formulation with a pseudospectral Dedalus ball basis, validate it with convergence and energy-balance tests, compare stationary and homothetic tail extraction for the cubic monopole, survey null-infinity rates for p=3,5,7 and ell=0,1,2, and present a tuned one-parameter family for the cubic monopole in which the leading e^{-tau} coefficient is cancelled and an e^{-2tau} tail is observed.","tokens_in":23439,"tokens_out":6241,"duration_ms":66824,"significance":"The proposed method addresses a real bottleneck in compactified tail computations: the radial steepening caused by observer-dependent decay rates. If the central claim holds, the homothetic formulation offers exponential efficiency in reaching late retarded times and removes the late-time gradient problem, which would be valuable for numerical relativity and for testing tail asymptotics. The paper's strengths include a clear coordinate construction, transparent balance laws, spectral convergence tests in both formulations, and numerical evidence (Fig. 4, Table I) that the method works for the cubic and quintic monopole and for higher modes. The tuned-monopole experiment is a genuinely interesting test of the 'generic but not universal' character of Rinne's conjecture. However, the central statement is analytically conditional: it relies on an assumed self-similar representation that is not proven for ell>=1, and the numerical verification of the uniform-in-radius property is currently limited to the cubic monopole. The p=7 rates also deviate by 5-6%, so the survey is only partially consistent with the conjectured rates.","major_comments":[{"comment":"The uniform exponential rate and the smooth-profile claim are derived from an assumed leading asymptotic representation chi_l = u^{-q}F_l(rho)+R_l with uniform remainder estimates, including u du R_l = O(u^{-q-delta} rho^{l+1}). The paper explicitly says 'we assume that a leading asymptotic representation exists.' For l=0, the cited small-data estimate (56) supplies the leading power, but I do not see that it supplies the uniform derivative-bound (65); for l>=1 neither the representation nor the remainder estimate is established. Eq. (70), which is the basis for the abstract and Conclusions, is therefore conditional. The authors should either prove or cite a proof of (63)-(65), or downgrade the presentation to a conditional/heuristic statement and mark the ell>=1 efficiency claim as numerically motivated rather than established.","section":"Sec. V C, Eqs. (63)-(65)"},{"comment":"The numerical support for the central claim is incomplete. Fig. 4 demonstrates the uniform decay rate only for the cubic monopole, and Table I reports fitted rates only at rho=1. No direct measurement is shown at several interior compactified radii for ell=1,2 or for p=5,7, so the key 'same rate at every fixed compactified radius' property is not verified for the higher multipoles that are central to the claimed advantage. In addition, the p=7 entries are 5.255-5.288 rather than the conjectured 5, i.e., 5-6% off; the manuscript should address whether this is a systematic error, an insufficiently long fitting window, or a genuine discrepancy.","section":"Sec. VI E, Table I and Fig. 4"},{"comment":"The blanket statement that the tail approaches the same decay rate at every compactified radius is stronger than the derivation. Eq. (70) holds only at radii where P_l(rho) is nonzero, and the fixed-radius rate (71) additionally requires K_l(0) to be nonzero. Neither condition is established for the semilinear higher-multipole solutions considered here. The abstract and Section VII should be reworded to include these qualifications, or the nonvanishing conditions should be verified, at least numerically for the cases presented.","section":"Eqs. (70)-(71) and Sec. VII"}],"minor_comments":[{"comment":"The text says 'medians over all 20 radii are 1.010 to the quoted precision, with a total spread of only 6e-6'; this phrasing is confusing because the top-right panel appears to show a larger spread over the plotted interval. Clarify whether the spread refers only to the late-time window and define what 'quoted precision' means.","section":"Sec. VI B, Fig. 4 caption"},{"comment":"The notation p=0 for the linear equation is nonstandard because the nonlinearity is written as phi^p; a reader may expect p=0 to denote a constant source. Define p=0 explicitly as the linear case in the text and caption.","section":"Sec. VI D, Fig. 6"},{"comment":"The fit model e^tau a00 = A1 + B e^{-tau} assumes the expansion (75), but no residual or goodness-of-fit statistic is reported for the tuned run. Since the claim of a clean e^{-2tau} window depends on this fit, reporting the residual would strengthen the evidence.","section":"Sec. VI F, Eq. (77)"},{"comment":"The conformal metric is identified with the static patch of de Sitter spacetime, but the Killing vector ∂_tau is not explicitly related to the de Sitter time orientation; a one-sentence clarification would help readers unfamiliar with horizon-fixing coordinates.","section":"Sec. IV B, Eq. (40)"},{"comment":"The signs in the source terms S_hom and S_stat are written as '±' with a note that the sign is chosen according to focusing or defocusing nonlinearity; since the paper consistently uses the focusing sign in the main text, it would be clearer to remove the ambiguity by stating the implemented sign explicitly.","section":"Appendix A, Eqs. (A1)-(A6)"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope for a numerical relativity / mathematical physics journal and the method is promising. The main risk is that the central claim is presented as established while Sec. V C explicitly marks the key input as an assumption. A major revision is appropriate: either supply a proof or citation for (63)-(65) or reframe the analytic part as a conditional mechanism, and add numerical checks of the uniform-in-radius rate for higher multipoles. The p=7 discrepancy and the missing interior-radius measurements should also be addressed before the work is accepted."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nShort version: this is a useful numerical methods paper, and the cubic monopole mechanism is demonstrated convincingly, but the paper's strongest advertised claim—uniform exponential decay at every fixed compactified radius for all multipoles—is conditional on a self-similar ansatz that is stated rather than proved, and the authors' own p=7 numbers do not fully reproduce the conjecture.\n\nWhat is actually new: the 3+1 pseudospectral implementation of homothetic hyperboloidal coordinates for semilinear wave tails, including balance laws, plus the numerical demonstration that in these coordinates the cubic monopole tail approaches a fixed smooth radial profile with the same exponential rate at all radii. The convergence tests and energy balance residuals (around 1e-9 to 1e-10) are solid, and the cost comparison—reaching u≈e^7 in 700 steps versus an estimated 1.1e5 stationary steps—is convincing. The authors also cite the mathematical literature for the homothetic construction and Rinne for the rate conjecture, so the credit allocation is honest. The tuned λ* cancellation is transparently a bisection, not a prediction; that is fine as long as it is labeled, and it is.\n\nSoft spots, in proportion. The uniform-rate argument in Sec. V C rests on the assumed representation (63)-(65), including the uniform derivative remainder bound. The paper says 'we assume that a leading asymptotic representation exists.' For l=0 the small-data estimate gives the leading power, but I do not see that it supplies the uniform derivative bound; for l>=1 neither part is proved. The numerical check of the uniform profile across radii is only the cubic monopole (Fig. 4); Table I reports rates only at ρ=1. So the claim that higher multipoles also enjoy the uniform profile is an ansatz, not a demonstrated result. The authors should either prove a weaker version for l>=1 or verify it numerically at several radii.\n\nSecond, the p=7 rates are 5.26-5.29 against the predicted 5, about 5-6% off. The abstract says the generic rates are reproduced, but this is not reproducing the p=7 rate. It may be a fitting-window or resolution effect, but without error bars or a multi-radius check we cannot tell. Code and data are promised, not provided; for a method paper that is a real gap.\n\nWho this is for: numerical relativists computing late-time tails, and people working on hyperboloidal methods. It deserves a serious referee. I would recommend major revision with release of code/data and either a proof or explicit conjecture for the higher-multipole uniform profile, plus a discussion of the p=7 discrepancy. The cubic mechanism is likely right and the method is worth having in the literature.","headline":"A useful numerical method for tail computations whose strongest advertised claim—uniform exponential rates for every multipole—rests on an unproved self-similar ansatz, and whose own p=7 rates do not fully reproduce the conjecture.","tokens_in":23970,"tokens_out":3872,"would_cite":true,"duration_ms":41439,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35L05","35B40","65M70"],"pacs":[],"model":"deepseek-v4-flash","headline":"Homothetic hyperboloidal coordinates turn semilinear wave tails into a fixed smooth profile with one decay rate at every compactified radius, removing the late-time gradient problem and reducing the step count from $O(U)$ to $O(\\log U)$.","keywords":["semilinear wave equations","late-time tails","homothetic hyperboloidal coordinates","self-similar decay","null infinity","tail decay rates","pseudospectral methods","Minkowski spacetime"],"falsifier":"Evolve a compactly supported $\\ell=1$, $p=3$ mode to $\\tau\\approx 10$ at high radial resolution and compare the fitted local exponent $-\\partial_\\tau\\ln|\\chi_1|$ at two fixed homothetic radii, say $\\rho=0.5$ and $\\rho=1$, over a window of width $\\Delta\\tau=1$; if the exponents differ beyond numerical error, or if the normalized profiles $e^{\\tilde q_1\\tau}\\chi_1(\\tau,\\rho)$ continue to change shape as $\\tau$ grows, the uniform-profile claim fails. Equivalently, compute the same tail in standard retarded coordinates and test whether $u^{q}\\chi_\\ell(u,\\rho)$ becomes independent of $u$ at fixed $\\rho$; a residual dependence decaying more slowly than $u^{-\\delta}$ would falsify the assumed self-similar representation.","tokens_in":22940,"feed_emoji":"🌊","tokens_out":13413,"duration_ms":123192,"temperature":0.7,"pith_summary":"Late-time wave tails decay more slowly along future null infinity than along timelike worldlines at fixed radius, so a compactified numerical grid must represent two different decay laws separated by an increasingly sharp transition. The paper argues that this steepening is an artifact of coordinates adapted to time translations rather than to the tail itself. Using homothetic hyperboloidal coordinates adapted to the scaling symmetry, the tail approaches a fixed smooth radial profile with the same exponential decay rate at every compactified radius; because homothetic time is the logarithm of retarded time, reaching very late retarded times costs only logarithmic numbers of steps. The authors demonstrate the mechanism with 3+1-dimensional pseudospectral evolutions, reproduce the conjectured generic decay rates for several powers and angular modes, and tune a one-parameter family of cubic monopole data to cancel the leading null-infinity coefficient and expose a faster subleading tail.","feed_headline":"Tail decay becomes one smooth profile in scaling-adapted coordinates","feed_subtitle":"One decay rate at every compactified radius cuts the cost of reaching late retarded times from linear to logarithmic.","key_machinery":"The central object is the homothetic hyperboloidal coordinate system $\\tau=\\ln((t^2-r^2)/(2Rt))$, $\\rho=r/t$ on the future cone, with conformal factor $\\Omega=1/t$; it is adapted to the dilatation operator $D=t\\partial_t+r\\partial_r$, the conformal metric is the static patch of de Sitter space, and future null infinity is fixed at $\\rho=1$ with $u|_{I^+}=Re^{\\tau}$. The mechanism is the assumed leading self-similar tail representation $\\chi_\\ell(u,\\rho)=u^{-q}F_\\ell(\\rho)+R_\\ell(u,\\rho)$ with origin-regular $F_\\ell=\\rho^{\\ell+1}K_\\ell$ and uniform remainder bounds; substituting $u=2Re^{\\tau}/(1+\\rho)$ turns this into $\\chi_\\ell(\\tau,\\rho)=e^{-q\\tau}P_\\ell(\\rho)+O(e^{-(q+\\delta)\\tau}\\rho^{\\ell+1})$, giving the same exponent at every fixed $\\rho$ and a smooth profile $Q_\\ell=P_\\ell/\\rho$ for the evolved conformal field. The fixed-radius decay $t^{-(q+\\ell+1)}$ follows from the same formula by taking $\\rho=r/t\\to 0$.","core_discovery":"The central claim is that an asymptotic radiative mode of a semilinear tail has the homothetic form $\\chi_\\ell(\\tau,\\rho)\\sim e^{-\\tilde q_\\ell\\tau}P_\\ell(\\rho)$ at every fixed $0<\\rho\\le 1$, with a time-independent radial profile and one common decay exponent across the compactified ball. In stationary hyperboloidal coordinates the same physics gives different exponents near null infinity and in the interior, producing an ever-narrowing transition layer. In homothetic coordinates the faster interior decay is not a competing exponent on the grid: it is recovered kinematically through the origin-regularity factor $\\rho^{\\ell+1}$ as $\\rho=r/t\\to 0$ along a fixed physical radius. Numerically, local logarithmic rates at 20 extraction radii agree across the compactified interval to about $10^{-5}$ in homothetic coordinates, while stationary runs show rates ranging from about 2 in the interior to about 1 at null infinity; fitted exponents for $p=3,5,7$ and $\\ell=0,1,2$ match the conjectured generic rates $\\tilde q_\\ell=\\max(p-2,\\ell+1)$.","pith_inferences":["A testable extension is to apply the same profile-collapse diagnostic to other self-similar late-time regimes: if normalized profiles at different $\\tau$ collapse onto one curve, logarithmic step counts should follow.","For black-hole spacetimes, where the mass breaks exact scale invariance, the likely route is a hybrid scheme matching stationary horizon-penetrating coordinates to an asymptotically homothetic exterior, which the paper identifies as future work.","The bisection tuning of the leading tail coefficient suggests a general numerical technique for exposing subleading asymptotic terms in any multipole, beyond the cubic monopole example shown here."],"forward_implications":["For tail-dominated evolutions, reaching a fixed retarded time $U$ at null infinity costs $O(\\log U)$ time steps in homothetic coordinates instead of $O(U)$ steps in stationary hyperboloidal coordinates.","The normalized late-time radial profile approaches a fixed shape, so a fixed spectral grid remains resolved indefinitely instead of requiring adaptive refinement near null infinity.","The generic null-infinity rates $\\tilde q_\\ell=\\max(p-2,\\ell+1)$ and the finite-radius rates $q_\\ell=\\tilde q_\\ell+\\ell+1$ are reproduced numerically for compactly supported data.","Tuning a one-parameter family of cubic monopole data can cancel the leading $e^{-\\tau}$ coefficient at null infinity, leaving a resolved $e^{-2\\tau}$ tail; generic-rate statements therefore hold on an open dense set, not for all smooth data.","Initial data with non-negligible support at null infinity decay as $q_{I^+}^{\\mathrm{nc}}=\\ell+1$ independent of the nonlinearity power, so tail-rate predictions must specify the data class."],"supporting_citations":[{"why":"proposes the higher-multipole decay-rate formulas and provides the stationary hyperboloidal tail results the paper reproduces and extends.","marker":"[29]"},{"why":"supplies the rigorous small-data uniform decay estimates for spherical tails that anchor the spherical rates.","marker":"[3]"},{"why":"introduces the self-similar homothetic coordinate construction for the cubic wave equation that the numerical formulation is built on.","marker":"[32]"},{"why":"identifies the late-time radial gradient steepening problem in stationary hyperboloidal compactified evolutions that motivates the new coordinates.","marker":"[14]"},{"why":"established the split between null-infinity and finite-radius decay rates in hyperboloidal numerical studies of the cubic wave equation.","marker":"[35]"},{"why":"provides the scri-fixing hyperboloidal compactification framework and energy-balance setup underlying both coordinate systems.","marker":"[15]"},{"why":"gives the late-time asymptotics for semilinear tails in the timelike cone used for the spherical rate comparison.","marker":"[47]"},{"why":"supplies the regular full-ball spectral formulation used to handle origin regularity in the 3+1 evolutions.","marker":"[45]"},{"why":"provides the spectral discretization library whose ball basis is used in the numerical implementation.","marker":"[53]"}],"fun_headline_variants":["Homothetic coordinates flatten tail decay profiles","One decay rate for all radii in tail evolution","Logarithmic runtime for late-time wave tails","Smooth tail profiles via scaling-adapted grids","Uniform tail decay cuts late-time cost"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes, rather than proves, that every mode has a leading self-similar tail representation $\\chi_\\ell(u,\\rho)=u^{-q}F_\\ell(\\rho)+R_\\ell(u,\\rho)$ with an origin-regular profile $F_\\ell=\\rho^{\\ell+1}K_\\ell$ and uniform remainder bounds; this is supported by small-data theory in the spherical case, but for higher multipoles it is exactly the unproved structure that produces the uniform exponential rate and the logarithmic cost, and the fixed-radius rate additionally needs $K_\\ell(0)\\neq 0$.","fun_headline_variants_meta":{"raw":{"variants":["Homothetic coordinates flatten tail decay profiles","One decay rate for all radii in tail evolution","Logarithmic runtime for late-time wave tails","Smooth tail profiles via scaling-adapted grids","Uniform tail decay cuts late-time cost"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000586,"raw_usage":{"total_tokens":2763,"prompt_tokens":965,"completion_tokens":1798,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":581,"completion_tokens_details":{"reasoning_tokens":1730}},"tokens_in":581,"tokens_out":1798,"duration_ms":13071,"temperature":1.0,"reasoning_tokens":1730,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T04:22:36.737574+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evolve a compactly supported $\\ell=1$, $p=3$ mode to $\\tau\\approx 10$ at high radial resolution and compare the fitted local exponent $-\\partial_\\tau\\ln|\\chi_1|$ at two fixed homothetic radii, say $\\rho=0.5$ and $\\rho=1$, over a window of width $\\Delta\\tau=1$; if the exponents differ beyond numerical error, or if the normalized profiles $e^{\\tilde q_1\\tau}\\chi_1(\\tau,\\rho)$ continue to change shape as $\\tau$ grows, the uniform-profile claim fails. Equivalently, compute the same tail in standard retarded coordinates and test whether $u^{q}\\chi_\\ell(u,\\rho)$ becomes independent of $u$ at fixed $\\rho$; a residual dependence decaying more slowly than $u^{-\\delta}$ would falsify the assumed self-similar representation.","supporting_citations":[{"cited_title":"In a direct implementation, regularity is obtained through cancellations between radial and angular terms","cited_arxiv_id":null,"evidence_quote":"supplies the rigorous small-data uniform decay estimates for spherical tails that anchor the spherical rates."},{"cited_title":"Rinne, Nonlinearity38, 105026 (2025), arXiv:2507.00674 [cs.NA]","cited_arxiv_id":null,"evidence_quote":"introduces the self-similar homothetic coordinate construction for the cubic wave equation that the numerical formulation is built on."},{"cited_title":"Nondispersive decay for the cubic wave equation","cited_arxiv_id":"1304.4135","evidence_quote":"established the split between null-infinity and finite-radius decay rates in hyperboloidal numerical studies of the cubic wave equation."}],"review_version":1}